Tensor tomography on surfaces (Q358819): Difference between revisions

Let \((M,g)\) be a compact oriented two-dimensional manifold with smooth boundary. Consider the geodesic ray transform acting on symmetric \(m\)-tensor fields on \(M\). In the case where the metric is Euclidean and \(m=0\), this transform is the so-called X-ray Radon transform, obtained by averaging functions along straight lines. In order to define the geodesic ray transform more generally, we require some notation. Given a symmetric covariant \(m\)-tensor field \(f=f_{i_1\cdots i_m}dx^{i_1} \otimes \cdots \otimes dx^{i_m}\) on \(M\), the corresponding function on the unit circle bundle \(SM\) is defined by \(f(x,v) = f_{i_1\cdots i_m} v^{i_1} \cdots v^{i_m}\). Here \(SM = \{(x,v) \in TM: |v|=1\}\). Geodesics going from \(\partial M\) into \(M\) are parametrized by the set \(\partial_+ (SM) = \{(x,v) \in SM: x \in \partial M, \langle v, \nu \rangle \leq 0\}\), where \(\nu\) is the outer unit normal vector to \(\partial M\). For \((x,v) \in SM\), \(t \mapsto \gamma(t,x,v)\) is the geodesic starting from \(x\) in the direction \(v\). It is assumed that \((M,g)\) is \textit{nontrapping}, namely that the time \(\tau(x,v)\) when the geodesic \(\gamma (t,x,v)\) exits \(M\) is finite for each \((x,v) \in SM\). The ray transform of \(f\) is then defined by \[ If(x,v) = \int_0^{\tau(x,v)} f(\phi_t(x,v))dt, \] where \((x,v) \in \partial_+(SM)\) and \(\phi_t\) denotes the geodesic flow of the Riemannian metric \(g\) acting on \(SM\). The paper under review considers the \(s\)-injectivity of the ray transform. Any tensor field \(f\) may be written uniquely as \(f=f^s + d_\sigma h\), where \(f^s\), the solenoidal part, is a symmetric \(m\)-tensor with zero divergence and in \(d_\sigma h\), the potential part, \(h\) is an \((m-1)\)-tensor with \(h|_{\partial M}=0\). \(I\) is said to be \(s\)-injective if \(I\) is injective on the set of solenoidal tensors. The paper under review first considers the case where \((M,g)\) is \textit{simple}; this ensures that it is nontrapping. The main result of the paper settles a long-standing question, in the two-dimensional case, on the \(s\)-injectivity of the ray transform: Let \((M,g)\) be a simple \(2D\) manifold and let \(m \geq 0\). If \(f\) is a smooth symmetric \(m\)-tensor field on \(M\) which satisfies \(If=0\), then \(f=d_\sigma h\) for some smooth symmetric \((m-1)\)-tensor field \(h\) on \(M\) with \(h|_{\partial M}=0\); moreover if \(m=0\) then \(f=0\). The paper gives two proofs of this result, one of which leads to a more general result. Let \(I_m\) denote the ray transforms acting on symmetric \(m\)-tensors; one can then use the natural \(L^2\) inner product to define an adjoint \(I_0^*\) of \(I_0\). The paper also proves the following: Let \((M,g)\) be a compact nontrapping surface with strictly convex smooth boundary. Suppose in addition that \(I_0\) and \(I_1\) are \(s\)-injective and that \(I_0^*\) is surjective. If \(f\) is a smooth symmetric \(m\)-tensor field on \(M\), \(m \geq 1\), which satisfies \(I_m f=0\), then \(f=d_\sigma h\) for some smooth symmetric \((m-1)\)-tensor field \(h\) on \(M\) with \(h|_{\partial M}=0\). This leads the authors to pose the following conjecture: Let \((M,g)\) be a compact nontrapping surface with strictly convex boundary. If \(f\) is a smooth symmetric \(m\)-tensor field on \(M\) which satisfies \(I f=0\), then \(f=d_\sigma h\) for some smooth symmetric \((m-1)\)-tensor field \(h\) on \(M\) with \(h|_{\partial M}=0\); moreover if \(m=0\) then \(f=0\).